Adds main script

main.py

@@ -0,0 +1,159 @@
+#!/usr/bin/env python3
+
+import pickle
+import pprint
+import time
+import wave
+
+import numpy as np
+import pyaudio
+import whisper
+from pynput import keyboard
+
+
+class SpeachToText:
+    def __init__(self, device_name) -> None:
+        self.p = pyaudio.PyAudio()
+        self.stream = None
+
+        # find the index of the specified device
+        device_index = None
+        for i in range(self.p.get_device_count()):
+            info = self.p.get_device_info_by_index(i)
+            if info['name'] == device_name:
+                device_index = info['index']
+                break
+
+        if device_index is None:
+            raise ValueError(f"Could not find audio device named {device_name}")
+
+        # open the audio stream
+        self.CHUNK = 512
+        FORMAT = pyaudio.paInt16
+        CHANNELS = 1
+        RATE = 16000
+
+        print()
+        # print ready in green
+        print("\033[92m{}\033[00m".format("ready!"))
+
+        self.stream = self.p.open(format=FORMAT,
+                                  channels=CHANNELS,
+                                  rate=RATE,
+                                  input_device_index=device_index,
+                                  input=True,
+                                  frames_per_buffer=self.CHUNK)
+
+        # setup whisper
+        self.model = whisper.load_model("base.en")
+
+        self.keyboard = keyboard.Controller()
+
+    # make destructor to close the stream and terminate the pyaudio instance
+    def __del__(self):
+        # if there is a stream open close it
+        if self.stream is not None:
+            self.stream.stop_stream()
+            self.stream.close()
+        if self.p is not None:
+            self.p.terminate()
+
+    # note that
+    def infer(self, frames):
+        # a np array containing the audio waveform
+        data = b''.join(frames)
+        data = np.frombuffer(data, dtype=np.int16)
+
+        data_to_transcribe = data.copy().flatten().astype(np.float32) / 32768.0
+
+        whisper_output = self.model.transcribe(data_to_transcribe, language="en")
+
+        import pprint
+
+        # pprint.pprint(result)
+        # set the output text and remove the leading space
+        output_text = whisper_output["text"][1:]
+        print(output_text)
+
+        self.keyboard.type(output_text)
+
+        return output_text
+
+    def run(self):
+        current_keys = set()
+
+        with keyboard.Listener(on_press=lambda key: current_keys.add(key), on_release=lambda key: current_keys.discard(key)) as listener:
+
+            # F16 key
+            activation_key = 269025095
+            activation_key_pressed = False
+
+            frames = []
+
+            try:
+                while True:
+
+                    if keyboard.KeyCode.from_vk(activation_key) in current_keys:
+                        if not activation_key_pressed:
+                            print()
+                            # print listening in cyan
+                            print("\033[96m{}\033[00m".format("listening..."))
+                            self.stream.start_stream()
+                            activation_key_pressed = True
+                        frames_to_read = self.stream.get_read_available()
+                        if frames_to_read > 0:
+                            data = self.stream.read(frames_to_read, exception_on_overflow=False)
+                            frames.append(data)
+                    elif activation_key_pressed:
+                        # process the frames after the activation key is released
+                        self.stream.stop_stream()
+                        self.infer(frames)
+                        frames = []
+                        activation_key_pressed = False
+
+            except KeyboardInterrupt:
+                # handle the Ctrl+C interrupt
+                print("Exiting...")
+
+
+def graph_frames_from_pyaudio(frames):
+    import matplotlib.pyplot as plt
+    import numpy as np
+
+    # get the audio data from the frames
+    data = b''.join(frames)
+    data = np.frombuffer(data, dtype=np.int16)
+
+    # plot the audio data
+    plt.plot(data)
+    plt.show()
+
+
+def get_devices():
+    p = pyaudio.PyAudio()
+
+    # list the devices
+    print()
+    for i in range(p.get_device_count()):
+        print()
+        pprint.pprint(p.get_device_info_by_index(i))
+
+
+def pickle_frames(frames):
+    data = b''.join(frames)
+    data = np.frombuffer(data, dtype=np.int16)
+    with open('frames.pickle', 'wb') as f:
+        pickle.dump(data, f)
+
+
+def unpickle_frames():
+    print("Loading frames from pickle...")
+    with open('frames.pickle', 'rb') as f:
+        data = pickle.load(f)
+    print("done loading")
+    return data
+
+
+if __name__ == '__main__':
+    speech_to_text = SpeachToText('')
+    speech_to_text.run()